Why most kids' coding courses stall

Nearly every parent we talk to has the same story. The kid was keen. They did a term of an after-school class, or a few weeks of an app that teaches loops by dragging blocks around. They got through the modules, and then they quietly stopped. The laptop went back to YouTube.

The usual diagnosis is that they lost interest. We think it's narrower than that: nothing they made was ever used by anyone. A finished module is just a finished module. Nobody plays it, nobody asks for a feature. Programming gets its hooks into you when you change something and a person on the other end reacts to it, and that loop never closed for them.

Kids who stick with it almost always have the project first and pick up the theory second, in whatever order the project demands. That's an awkward thing to say to someone who has just paid for a course, but it matches what we see in adults. We've hired developers who came up through computer science degrees and developers who came up through modding games at fourteen. The second group debugs faster.

A worked example: A-Lotl Games

A-Lotl Games is a free browser games site built by Emmett, who is twelve. He's the son of Code Workshop's founder, which is the only reason we can tell you what actually happened rather than what people generally claim happens.

There are more than twenty original games on it now. They run in a browser tab with no download and no app store, which sounds like a technical detail but is really the whole distribution strategy: a kid at school can open a link on a locked-down laptop and play. The games are also unmistakably Australian in a way you don't get from following a tutorial.

The part that surprised us

Plenty of kids make a game. Fewer get to everything that has to surround a game before strangers can use it, and that's where the real learning showed up.

A-Lotl Games has accounts and logins. It has leaderboards that persist between sessions. There's an in-site shop, a dev console, and a set of Minecraft mods. Several of the games are properly online: Dungeon Quest runs a co-op session for up to ten players, and Walkabout Tag puts everyone in a shared world in real time.

The multiplayer runs peer-to-peer over WebRTC, so players connect to each other directly instead of going through a game server somebody has to pay for. Working out that this was the answer, and then getting it to work, is not a beginner problem. He needed multiplayer, he had no budget for servers, and he found the approach that doesn't need one. Most of our job looks like that.

None of it sits on a game engine. It's hand-written JavaScript drawing to a canvas, which is the same set of skills behind a lot of what we build for clients. We wrote a longer piece on that in building a browser drawing engine from scratch.

Where AI fits

Emmett is self-taught, and he's leaned on AI to get there. That's the question every parent is actually asking in 2026, and most of the available answers are either panic or a sales pitch.

The distinction that matters is explain versus finish. Asking an AI why something says undefined, or what an error means, is using a tutor with unlimited patience. That's what a twelve-year-old working alone at 9pm needs, and it's what a parent who doesn't code can't provide. Asking it to build a racing game and pasting the answer in teaches nothing, and it falls over the first time the code needs changing.

What AI removed is the thing that used to kill kids' projects: three days stuck on a stray semicolon with nobody to ask. It didn't remove the need to understand your own code. If you want a test, ask your kid to explain what a chunk of their code does. If they can, it's theirs, however it got typed. If they can't, that's a problem heading their way regardless.

How to help without taking over

What you actually need to start

Almost nothing, which is the part people miss. A text editor, a browser, and somewhere to put the files, all of it free. The browser your kid already has can draw graphics, play sound, save progress and open direct connections to other players. There's no engine to license and no store approval to wait on.

If they want to see what someone their age did with exactly that, A-Lotl Games is sitting there in public, including a Fun Code section of copy-and-paste console tricks Emmett put together for other kids. Start by copying an idea and making it worse.

Why a software company is writing about this

Partly because it's our kid and we're allowed to be pleased about it. It's also the clearest illustration we have of how we think about building software generally.

The instinct we run into constantly in business software is to plan the whole system, specify it completely, and build all of it before anyone touches anything. It rarely works, for the same reason the coding course rarely works. Nothing gets used, so nothing gets learned, and by the time real people see it the assumptions baked in six months ago have quietly gone stale. Build a small real thing, watch someone use it, then fix what they hit. That holds whether you're twelve and building a game about magpies or running a business that needs software fitting how it actually operates.

We're an intentionally small team in Bowral, and that's roughly how we work with everyone.